6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-12-13 04:30:23 +00:00
Code Issues Projects Releases Wiki Activity
release_29
llvm-6502/test/CodeGen/X86/memcpy-2.ll

168 lines
4.0 KiB
LLVM
Raw Permalink Normal View History

Add -mcpu to memcpy / memset tests to ensure they behave the same on all hosts / targets. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100101 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 08:25:26 +00:00
; RUN: llc < %s -mattr=+sse2 -mtriple=i686-apple-darwin -mcpu=core2 | FileCheck %s -check-prefix=SSE2
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; RUN: llc < %s -mattr=+sse,-sse2 -mtriple=i686-apple-darwin -mcpu=core2 | FileCheck %s -check-prefix=SSE1
Add -mcpu to memcpy / memset tests to ensure they behave the same on all hosts / targets. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100101 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 08:25:26 +00:00
; RUN: llc < %s -mattr=-sse -mtriple=i686-apple-darwin -mcpu=core2 | FileCheck %s -check-prefix=NOSSE
In 64-bit mode, use i64 to lower memcpy / memset instead of f64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 20:27:45 +00:00
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=core2 | FileCheck %s -check-prefix=X86-64
Add a couple of test cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51441 91177308-0d34-0410-b5e6-96231b3b80d8
2008-05-22 21:19:19 +00:00
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
@.str = internal constant [25 x i8] c"image\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00\00"
@.str2 = internal constant [30 x i8] c"xxxxxxxxxxxxxxxxxxxxxxxxxxxxx\00", align 4
Add a couple of test cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51441 91177308-0d34-0410-b5e6-96231b3b80d8
2008-05-22 21:19:19 +00:00
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
define void @t1(i32 %argc, i8** %argv) nounwind {
Add a couple of test cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51441 91177308-0d34-0410-b5e6-96231b3b80d8
2008-05-22 21:19:19 +00:00
entry:
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
; SSE2: t1:
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
; SSE2: movaps _.str, %xmm0
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
; SSE2: movaps %xmm0
; SSE2: movb $0
; SSE2: movl $0
; SSE2: movl $0
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; SSE1: t1:
; SSE1: movaps _.str, %xmm0
; SSE1: movaps %xmm0
; SSE1: movb $0
; SSE1: movl $0
; SSE1: movl $0
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
; NOSSE: t1:
; NOSSE: movb $0
; NOSSE: movl $0
; NOSSE: movl $0
; NOSSE: movl $0
; NOSSE: movl $0
; NOSSE: movl $101
; NOSSE: movl $1734438249
In 64-bit mode, use i64 to lower memcpy / memset instead of f64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 20:27:45 +00:00
; X86-64: t1:
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
; X86-64: movaps _.str(%rip), %xmm0
In 64-bit mode, use i64 to lower memcpy / memset instead of f64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 20:27:45 +00:00
; X86-64: movaps %xmm0
; X86-64: movb $0
; X86-64: movq $0
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
%tmp1 = alloca [25 x i8]
%tmp2 = bitcast [25 x i8]* %tmp1 to i8*
call void @llvm.memcpy.i32( i8* %tmp2, i8* getelementptr ([25 x i8]* @.str, i32 0, i32 0), i32 25, i32 1 ) nounwind
unreachable
Add a couple of test cases. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51441 91177308-0d34-0410-b5e6-96231b3b80d8
2008-05-22 21:19:19 +00:00
}
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
;rdar://7774704
%struct.s0 = type { [2 x double] }
define void @t2(%struct.s0* nocapture %a, %struct.s0* nocapture %b) nounwind ssp {
entry:
; SSE2: t2:
; SSE2: movaps (%eax), %xmm0
; SSE2: movaps %xmm0, (%eax)
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; SSE1: t2:
; SSE1: movaps (%eax), %xmm0
; SSE1: movaps %xmm0, (%eax)
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
; NOSSE: t2:
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
In 64-bit mode, use i64 to lower memcpy / memset instead of f64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 20:27:45 +00:00
; X86-64: t2:
; X86-64: movaps (%rsi), %xmm0
; X86-64: movaps %xmm0, (%rdi)
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
%tmp2 = bitcast %struct.s0* %a to i8* ; <i8*> [#uses=1]
%tmp3 = bitcast %struct.s0* %b to i8* ; <i8*> [#uses=1]
tail call void @llvm.memcpy.i32(i8* %tmp2, i8* %tmp3, i32 16, i32 16)
ret void
}
define void @t3(%struct.s0* nocapture %a, %struct.s0* nocapture %b) nounwind ssp {
entry:
; SSE2: t3:
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; SSE2: movsd (%eax), %xmm0
; SSE2: movsd 8(%eax), %xmm1
; SSE2: movsd %xmm1, 8(%eax)
; SSE2: movsd %xmm0, (%eax)
; SSE1: t3:
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
; SSE1: movl
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
; NOSSE: t3:
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
; NOSSE: movl
In 64-bit mode, use i64 to lower memcpy / memset instead of f64. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100137 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 20:27:45 +00:00
; X86-64: t3:
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; X86-64: movq (%rsi), %rax
; X86-64: movq 8(%rsi), %rcx
; X86-64: movq %rcx, 8(%rdi)
; X86-64: movq %rax, (%rdi)
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
%tmp2 = bitcast %struct.s0* %a to i8* ; <i8*> [#uses=1]
%tmp3 = bitcast %struct.s0* %b to i8* ; <i8*> [#uses=1]
tail call void @llvm.memcpy.i32(i8* %tmp2, i8* %tmp3, i32 16, i32 8)
ret void
}
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
define void @t4() nounwind {
entry:
; SSE2: t4:
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; SSE2: movw $120
; SSE2: movl $2021161080
; SSE2: movl $2021161080
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
; SSE2: movl $2021161080
; SSE2: movl $2021161080
; SSE2: movl $2021161080
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; SSE2: movl $2021161080
; SSE2: movl $2021161080
; SSE1: t4:
; SSE1: movw $120
; SSE1: movl $2021161080
; SSE1: movl $2021161080
; SSE1: movl $2021161080
; SSE1: movl $2021161080
; SSE1: movl $2021161080
; SSE1: movl $2021161080
; SSE1: movl $2021161080
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
; NOSSE: t4:
; NOSSE: movw $120
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; NOSSE: movl $2021161080
; X86-64: t4:
; X86-64: movabsq $8680820740569200760, %rax
; X86-64: movq %rax
Revert r122955. It seems using movups to lower memcpy can cause massive regression (even on Nehalem) in edge cases. I also didn't see any real performance benefit. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@123015 91177308-0d34-0410-b5e6-96231b3b80d8
2011-01-07 19:35:30 +00:00
; X86-64: movq %rax
; X86-64: movq %rax
Avoid using f64 to lower memcpy from constant string. It's cheaper to use i32 store of immediates. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100751 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-08 07:37:57 +00:00
; X86-64: movw $120
; X86-64: movl $2021161080
%tmp1 = alloca [30 x i8]
%tmp2 = bitcast [30 x i8]* %tmp1 to i8*
call void @llvm.memcpy.i32(i8* %tmp2, i8* getelementptr inbounds ([30 x i8]* @.str2, i32 0, i32 0), i32 30, i32 1)
unreachable
}
Fix sdisel memcpy, memset, memmove lowering: 1. Makes it possible to lower with floating point loads and stores. 2. Avoid unaligned loads / stores unless it's fast. 3. Fix some memcpy lowering logic bug related to when to optimize a load from constant string into a constant. 4. Adjust x86 memcpy lowering threshold to make it more sane. 5. Fix x86 target hook so it uses vector and floating point memory ops more effectively. rdar://7774704 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@100090 91177308-0d34-0410-b5e6-96231b3b80d8
2010-04-01 06:04:33 +00:00
declare void @llvm.memcpy.i32(i8* nocapture, i8* nocapture, i32, i32) nounwind
Reference in New Issue Copy Permalink